1. Field of the Invention
The present invention generally relates to image processing apparatuses and, more particularly, to an image processing apparatus, an image forming apparatus and an image forming method that detect a kind of images represented by binary image data, which is binarized according to an error diffusion method, so as to apply an appropriate image processing method to the binary image data in accordance with the result of detection.
2. Description of the Related Art
A document image comprises a character image, a drawing image including lines and curves, a character image or a drawing image provided with screening, a photograph image, or a combination of the aforementioned images. A copy machine that makes a copy of such an original document image converts the document image into digital imaged data, and applies various correction processes to the digital image data in accordance with the contents of the document image so as to reproduce a copy image closer to the original document image.
For example, a correction process that gives weight to sharpness is suitable for a document having contents such as a drawing consisting of only characters and lines. The document image consists of only a black part (bit information=“1111”) corresponding to the characters and lines and a while part (bit information=“0000”) where no character or line is drawn. Accordingly, a correction process is applied to well-define the black part and the white part (a correction process that gives weight to sharpness). According to such a correction process being applied to image data, a copy image having a well-defined contour of a character can be reproduced.
On the other hand, a correction process that gives weight to gradation is suitable for a correction to an original document having contents such as a photograph, a color picture, etc. If, for example, an original document is a photograph of sixteen gradation levels of black and white, the original document is composed of pixels of intense black (bit information=“1111”), stark white (“0000”), light gray (“0010”) and dark gray (“1100”). Since the bit information of consecutive pixels changes smoothly in general, that is, there is no sharp change in the bit information between adjacent pixels, it is necessary to apply a smooth correction (a correction that gives weight to gradation).
As appreciated from the above explanation, the correction that gives weight to sharpness and the correction that gives weight to gradation are conflict with each other. For example, if the correction that gives weight to sharpness is applied to image data of a photographic image, a reproduced image according to the corrected image data may become a so-called jugged reproduction image that lacks spatial effect. On the other hand, if the correction that gives weight to gradation is applied to a character or line drawing, a reproduced image according to the corrected image data may become a so-called blurry reproduction image that has unclear outlines.
Thus, even if an original document image contains both a character image and a photographic image, a reproduced image closer to the original document image can be obtained by applying a suitable image processing in accordance with portions of the original document image corresponding to the character image and the photographic image. For example, there is suggested, as a conventional technique, a technique with respect to a method of discriminating a dot-photograph, which discriminates a dot-photograph area and a non-dot-photograph area (an area of characters or line drawing) in an input image in accordance with a result of measurement of a signal level difference between two consecutive pixels or a change in the sign (plus or minus) of the pixels with respect to both a main scanning direction and a sub-scanning direction (refer to Japanese Patent Publication No. 6-66884).
Additionally, there is suggested a technique with respect to a digital image reading apparatus, which can prevent a reproduced image from being degraded by applying either an image processing that gives priority to resolution or an image processing that gives priority to gradation in accordance with a result of determination as to whether a target pixel in input digital image data is in a character area or in a picture area (refer to Japanese Laid-Open Patent Application No. 9-233323).
Further, there is suggested a technique with respect to an image processing apparatus, and image processing method and an image forming apparatus, which divides an input image into a plurality of portions and determined whether or not each divided portion is a specific portion (such as a photograph, etc.) so as to perform an image processing that gives weight to gradation when each divided portion is the specific portion and perform an image processing that gives weight to sharpness when each divided portion is not the specific portion (refer to Japanese Laid-Open Patent Application No. 2003-467719).
The method of detecting an image portion according to the above-mentioned technique is suitable for detecting an image portion from image data before being converted into binary image data, that is, suitable for detecting an image portion from multi-value image data. However, the method of detecting an image portion according to the conventional technique is not suitable for the detection performed on the binary image data since the binary image to which an error diffusion process has been applied is generated by converting multi-gradation-level image into pseudo-halftone data, which contains black dots and white dots so as to represent various gray colors by changing a density of the black dots relative to the white dots. Therefore, in order to detect an image portion from image data to be transmitted to a plotter (write unit), it is necessary to detect an image portion from image data after the binarizing process.
Moreover, the detection method according to the conventional technique, a determination process of an image portion is performed in both the main-scanning direction and the sub-scanning direction. That is, since an amount of computation is about twice, a memory area must be increased accordingly. Moreover, since the multi-value image contains bit information from “0000” to “1111” in the case of sixteen gradation levels while the binary image contains bit information of “0” and “1”, the processing of the multi-value image requires a larger memory area than the processing of the binary image. Consequently, the processing of the multi-value image requires a large amount of computation, which results in an increase in the size and cost of the apparatus that performs the image processing. Thus, it is desirous to provide a relatively inexpensive apparatus by simplifying the structure of the apparatus while maintaining reproducibility of a reproduced image.
Moreover, in the conventional technique, a smoothing process is applied to the whole binary image so that the smoothing process matches only a character/line portion or matches only a non-character/line portion (photograph, etc.) or matches an intermediate between the two. If a strong smoothing that gives weight to the character/line portion is applied, there is no jaggy occurs in characters and lines, which results in a copy image having clear outlines, but a photograph portion or the like turns into a copy image having poor reproducibility. On the contrary, if a gentle smoothing that gives weight to the non-character/line portion is applied or no smoothing is applied, excellent reproducibility is obtainable for the non-character/line image, but reproducibility of the non-character/line image is poor. Additionally, the smoothing process for the intermediate between the character/line image and the non-character/line image may reproduce a copy image, which is insufficient for both the character/line image and the non-character/line image. Thus, in order to improve the reproducibility of a copy image, it is necessary to perform an image processing according to each image portion of the original document.